Advanced voice mode inferior to Legacy mode

I’m just sharing some feedback as a longtime user of the classic ChatGPT voice mode. I believe it’s still far superior to the recent Advanced Mode for several reasons. My goal in writing this is to urge OpenAI to preserve access to the legacy mode indefinitely, so users continue to have a choice. Here are several reasons why I feel this way:

Audio Quality

Most notably, the audio quality—specifically the fidelity of the voice—is seriously degraded in Advanced Mode. I imagine this was a compromise to devote some of that processing power to its expressivity and other features, but it feels like talking over a really bad, old-school cell phone connection.

Performative Voices

These voices sound exceedingly performative, disingenuous, and frankly irritating. It’s a cute little party trick for the first five minutes, but then I just find myself switching back to the old mode, where things actually feel more organic.

Restrictive Guardrails

I’ve seen this come up in other posts, but it’s worth underscoring—the guardrails seem incredibly limiting in Advanced Mode, even for conversation topics that are seemingly perfectly harmless.

Superficiality

There’s a lack of depth or nuance in Advanced Mode conversations. The bot seems to focus more on its verbal prowess than the actual heart of the conversation.

A Note About the Legacy Sky Voice

For what it’s worth, I still greatly miss the legacy Sky voice and hope that, once all the legal rigmarole is sorted out, it will be re-added. Let’s not forget: the company said they were pausing the voice, not discontinuing it. As a long-time paying user, I’m still expecting the Sky voice to be restored—or for the company to definitively tell us that it won’t be. To this day, it remains the very best of the original voices, with Juniper being a close runner-up.

Thanks.

1 Like

I find the audio quality, especially the microphone on Android is completely unacceptable. High constantly have to stop and go back to voice typing input which I find is effectively the same thing. Interruptions are constant, and it just doesn’t understand me nearly as well as whisper mode or voice typing using Google.